Demand Forecasting System
A very common issue within the production and packaging industry is the inability to accurately forecast demand, this single metric has a lot of weightage within the profits of companies within this industry. Even though customers of these companies provide estimations of the amount that will be purchased (Written in agreement), in reality there is always a variance in the amount actually purchased.
This issue creates the following problems:
- Overrun of raw materials (Excess costs, excess inventory, excess management time)
- Under purchase of raw materials (Unable to meet customer expectations)
- Estimations dictate the price per piece, a lot of the time revenue is missed out due to agreed volumes not actually being purchased by the customer.
- Forecasts are manually created and need to be recreated very frequently. The manual creation of these forecasts takes around 2 hours per customer.
Before our client approached us to help with this issue, they manually created forecasts that took around 2 hours per customer to create, and needed to be recreated very frequently for it to be useful.
Our client wanted to address this issue by increasing the accuracy of forecasting demand and also reduce manual effort.
The amount of time it to create these forecasts decreased to less than 60 seconds as all the data automatically came into the forecasting system and was automatically computed.
Using historical data and other sources of data, we were able to produce an algorithm that increased the accuracy for the forecasts of demand for each customer. The system has helped the company increase productivity, decreases costs, and enhance customer experience.
For this particular project, we first worked on increasing the accuracy of a single customer before moving on to others. Due to the nature of the objective, we had to create a seperate algorithm for each customer our client serves. Since it wasn’t a lot of rework, the subsequent algorithms were created fairly quickly, allowing for a ROI within 12 months.